CN107678433B - Loading and unloading equipment scheduling method considering AGV collision avoidance - Google Patents
Loading and unloading equipment scheduling method considering AGV collision avoidance Download PDFInfo
- Publication number
- CN107678433B CN107678433B CN201710984769.5A CN201710984769A CN107678433B CN 107678433 B CN107678433 B CN 107678433B CN 201710984769 A CN201710984769 A CN 201710984769A CN 107678433 B CN107678433 B CN 107678433B
- Authority
- CN
- China
- Prior art keywords
- task
- agv
- time
- container
- tasks
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 238000000034 method Methods 0.000 title claims abstract description 36
- 238000005457 optimization Methods 0.000 claims abstract description 15
- 238000012544 monitoring process Methods 0.000 claims abstract description 10
- 230000008569 process Effects 0.000 claims description 17
- 230000001133 acceleration Effects 0.000 claims description 11
- 230000003068 static effect Effects 0.000 claims description 11
- 230000003993 interaction Effects 0.000 claims description 8
- 230000007704 transition Effects 0.000 claims description 6
- 238000012545 processing Methods 0.000 claims description 4
- 230000001174 ascending effect Effects 0.000 claims description 3
- 238000012546 transfer Methods 0.000 claims description 3
- 238000010586 diagram Methods 0.000 description 9
- 238000013507 mapping Methods 0.000 description 3
- 238000011160 research Methods 0.000 description 2
- 238000013459 approach Methods 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 150000001875 compounds Chemical class 0.000 description 1
- 238000000354 decomposition reaction Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G05—CONTROLLING; REGULATING
- G05D—SYSTEMS FOR CONTROLLING OR REGULATING NON-ELECTRIC VARIABLES
- G05D1/00—Control of position, course or altitude of land, water, air, or space vehicles, e.g. automatic pilot
- G05D1/02—Control of position or course in two dimensions
- G05D1/021—Control of position or course in two dimensions specially adapted to land vehicles
- G05D1/0212—Control of position or course in two dimensions specially adapted to land vehicles with means for defining a desired trajectory
- G05D1/0214—Control of position or course in two dimensions specially adapted to land vehicles with means for defining a desired trajectory in accordance with safety or protection criteria, e.g. avoiding hazardous areas
Abstract
The invention provides a dispatching method of loading and unloading equipment considering AGV collision avoidance. Establishing a dynamic model and two types of collision avoidance models of container terminal loading and unloading equipment; dynamically regarding discrete events of equipment as a scheduling problem of a mixed flow shop, and then scheduling all container tasks of the mixed flow shop by adopting an integral graph sequence method; and finally, under the condition of considering two collision avoidance situations, according to an introduced layered control structure, sending each generated operation time window to a stage controller by a state monitoring controller, distributing each operation time window to specific equipment by the stage controller, and updating the operation time of the AGV according to the minimum time control problem so as to realize collision-free scheduling of all tasks of the AGV. The scheduling method considering AGV collision avoidance and loading and unloading equipment simultaneously can meet the optimization requirement of scheduling and managing problems of a large-scale automatic container terminal.
Description
Technical Field
The invention relates to the field of transportation of an automatic container terminal, in particular to the field of dispatching of loading and unloading equipment of the automatic container terminal, and particularly relates to a method for dispatching the loading and unloading equipment by considering AGV collision avoidance.
Background
The problem of AGV trajectory planning in container terminals has received a lot of attention from researchers due to the close connection between devices and the complexity of the environment. At present, the dispatching of container terminal loading and unloading equipment usually ignores the actual trajectory planning problem of the equipment. In addition, in the research of container terminals, the devices are generally assumed to work without interference, and the scheduling scheme does not include interference caused by collision avoidance. In addition, some researchers have studied the traditional three-stage hybrid flow shop of an automated container terminal to consider only the single quay crane case, whereas in the actual terminal environment, the job usually involves multiple quay bridges, multiple AGVs and multiple bridges. In addition, the research does not comprehensively consider the interaction between the collision avoidance constraint of the AGV and the dispatching of the loading and unloading equipment, which can influence the dispatching optimization performance of the loading and unloading equipment at the container terminal; based on the method, the invention provides a loading and unloading equipment scheduling method considering AGV collision avoidance to optimize the problem of collision-free scheduling of a shore bridge, the AGV and a field bridge in the process of transporting containers.
Disclosure of Invention
The invention provides a loading and unloading equipment scheduling method considering AGV collision avoidance aiming at the problem of collision-free scheduling of loading and unloading equipment of an automatic container terminal, so as to realize collision-free scheduling of a terminal shore bridge, the AGV and a yard bridge. Firstly, establishing a dynamic model of a container terminal loading and unloading device and two types of collision avoidance models, wherein the dynamic model of a transport container can be represented by the combination of high-level discrete event dynamics and low-level continuous time dynamics, the scheduling of the loading and unloading device is related to the high-level discrete event dynamics, and the collision-free track planning is related to the low-level continuous time dynamics; dynamically regarding discrete events of equipment as a scheduling problem of a mixed flow shop, and then scheduling all container tasks of the mixed flow shop by adopting an integral graph sequence method; and finally, introducing a hierarchical control structure, determining the task sequence of each stage and minimizing the completion time of all tasks of each stage by a state monitoring controller under the condition of considering two collision avoidance constraints, sending the generated operation time windows to a stage controller, distributing the operation time windows to specific equipment by the stage controller, and updating the operation time of the AGV according to the minimum time control problem to realize collision-free scheduling of all tasks of the AGV.
Scheduling, as considered in this patent, includes determining the order of tasks for each particular device and the operating time window for each task; the scheduling is determined by dealing with a hybrid flow shop scheduling problem and an AGV collision avoidance problem. Two types of collision avoidance are considered: a stationary collision, which refers to the collision of the AGVs with stationary obstacles and associated equipment in front of the delivery point of the stacking area, and a moving collision, which refers to the collision of different AGVs with each other during the transportation of the container. The method for dispatching the loading and unloading equipment mainly comprises the following steps:
step one, establishing a model of three-stage equipment
In the three stage flow shop, a task is defined as a complete process of transporting a container from a ship to a designated location in a heap.
(1) Stage 1: multiple QCs
(2) And (2) stage: multiple AGV
(3) And (3) stage: multiple ASCs
Assuming that there are N containers to be transported from ship to heap, let Φ be the set of tasks (| Φ | ═ N), introduce two virtual task points 0 and N +1, and then let Φ be defined1Φ ∪ {0} and Φ2Φ ∪ { N +1 }. the time constraint for a particular device is as follows:
xij1 indicates that in the first phase, the container task i completes earlier than task j, otherwise, xij=0;
aiIs the completion time of container task i in the first phase; biIs the completion time of the container task i in the second stage; c. CiIs the completion time of the container task i in the third phase; diIs the start time for the transition from task i to task j in the second phase; e.g. of the typeiAGV arrival time at which container i is ready to be transported in the first stage;is thatCompletion time of h1∈{1,2,3},h2∈{1,2};Is the transfer time of container i to container j handled by the designated AGV in the second phase; r is a large positive number; at is the reserved time for two containers to be loaded and unloaded successively by the same ASC.
Inequalities (1) and (4) represent the first task to initialize the QC and AGV, respectively. Inequality (2) represents a relationship between task i and task j processed by QC; inequality (3) ensures that at any moment, the joint point before each QC has at most one AGV; inequalities (5) and (6) represent the start of the second phase after task i is completed in the first phase; inequalities (7) and (8) indicate that task i is started in the third phase after completion of the second phase; inequality (9) represents the relationship between container task j and task i handled by ASC; inequality (10) indicates that the reserved time delta t between two successive tasks is ensured under the constraint of only one AGV at most at the front junction of each heap at any moment; inequalities (11) and (12) represent the transitions of container tasks i and j transported by the AGV.
Because collision avoidance is considered, decision variables of various tasks cannot be determined simultaneously, and a graphical method can be adopted to represent the three-stage hybrid flow shop scheduling, wherein each graphical representation processes corresponding container tasks in a specific order at a specific stage; integer variable xij,yijAnd zijRespectively defining the sequence of the container tasks processed by specific equipment in the three stages; x is the number ofij,yijAnd zijThe values of (a) may be represented by a graph of the stages;
the graphic sequence S represents the scheduling sequence of all container tasks to be executed in a specific stage of the hybrid flow shop; for phase 1, defineWherein task j is executed after task i; for task i and task j, if task i and task j are accessed sequentially and both task i and task j are executed by a particular device in the graph, then xij1, otherwise xij0; stages 2 and 3 are defined similarly;
in order to schedule all container tasks of the mixed flow workshop in sequence, an overall graph sequence of each stage of the three stages is needed, and the index number of the sequence gives the operation sequence of the tasks; therefore, consider introducing a continuous variable bi(i e Φ), each task of the overall graph sequence being related to the values of these variables; it will be apparent that the variable values may be arranged in an ascending order to determine the job order of the tasks. And in the three-stage hybrid plant mentioned in step one, if there is R (1-x)ij)+bj>biThen, the sequence S {.. ·, i., j. } (b) }j≥biI, j e Φ, i ≠ j) is a feasible whole graph sequence. The reason is as follows:
consider an approximationBundles (7), (8) and (10) having R (1-z)ij)+bj>dj≥biSo that R (1-z)ij)+bj>bi. Because task i and task j are accessed sequentially for a particular AGV, R (1-y) may be obtainedij)+bj>bi. Thus, for the sequence S {.. ·, i., j. } (b)j≥bi,i,j∈Φ,i≠j),R(1-xij)+bj>bi,R(1-yij)+bj>biAnd R (1-z)ij)+bj>bi. This means that the sequence S is a feasible overall pattern sequence.
Step two, establishing an AGV and a collision avoidance model
1) Establishing AGV dynamic model
In the formula, AGVpIs at a position ofSpeed isAcceleration of a vehicle(up(k)∈R2),I2Is a 2 × 2 unit array, Δ t is the time interval;
in the formula umaxAnd vmaxUpper limit values of acceleration and velocity, respectively, and M is an arbitrary number.
2) Establishing static collision avoidance constraint model
The rectangular area may be represented by the lower left corner coordinate(s)low,x,slow,y) And coordinates of upper right corner(s)high,x,shigh,y) To represent; to avoid the static obstacle area, the position of AGVi must be outside the rectangular area; therefore, the constraint conditions for satisfying the collision avoidance requirement are as follows:
with binary variables introduced, the constraint (18) can be rewritten to the form of a standard optimization problem as follows:
wherein R is a large positive real number, bin,τ(k) Is a binary variable; constraints (19) and (20) ensure that at least one of equations (18) is true;
3) establishing a moving collision avoidance constraint model
In each time interval, except for the AGVp1And AGVp2Are different from each other, a minimum safety distance (2d) is maintained between them; the collision avoidance constraint can be written in the form of:
p2∈[1,...,nagv],p1≠p2in the formula (I), the compound is shown in the specification,andp of AGV in k1And p2The position of (a); introducing a binary variable, the constraint (21) can be rewritten as follows:
in the formula, bm,τIs a binary variable, R is a sufficiently large positive real number, and equations (20) and (21) ensure that equation (19) holds;
step three, determining a hierarchical control structure
According to the hierarchical control structure, the collision-free scheduling of all tasks is planned by the whole graph sequence, and the whole graph sequence simultaneously solves the problem of collision-free track planning and the scheduling of loading and unloading equipment; according to the overall graph sequence, sequentially completing collision-free scheduling of all tasks one by one; for a group of tasks with an operation sequence to be determined, the first planned task is used for determining the relevant collision-free track of the task in the phase 2 and the relevant operation time of the task in the phase 1 and the phase 3 controlled by a high-level controller; after the first task is completed, the high-level controller updates the discrete event dynamic and determines the operation sequence and the operation time of the rest tasks again; this process is repeated until the scheduling of all tasks is completed;
1) state monitoring controller
The state monitoring controller is used for determining the task sequence of each stage and minimizing the completion time of all tasks of each stage. When the QC and the AGV can be scheduled at any time, the completion time is defined as the completion time of the last ship leaving task. I.e., max { a }1,...,an,e1,...enI.e., | f | ceiling∞,||·||∞Denotes an infinite norm, wherein f ═ a1,a2,...,an,e1,...,en]T。
In order to determine the task order of all stages of the mixed flow shop problem, the theorem needs to add additional constraints; the following definitions are first made:
a=[a1,a2,...,an]T
b=[b1,b2,...,bn]T
c=[c1,c2,...,cn]T
d=[d1,d2,...,dn]T
e=[e1,e2,...,en]T
the scheduling problem model can be written as:
constraint conditions are as follows:
and (1) - (12) and associated equation constraints;
andare interdependent betweenAndstart time between transitionsOrDependent on the interaction between the devices performing task i; this interaction is implemented by a high-level controller that solves the following optimization problem:
constraint conditions are as follows: (1) (12) and associated equality constraints; the optimization problem is a linear programming problem, so that the time of the hybrid flow shop can be quickly adjusted;
2) phase controller
Considering the consistency of the dynamic models of all AGVs in stage 2, an AGV set is defined: Ψagv={1,2,...,nagv}
fagv:Φ→Ψagv, (25)
Wherein f isagvIs to map the task set Φ to the AGV set ΨagvA function of (a); f. ofagv(i) Assigned to task iThe AGV of (1); the time window assigned to AGV task i is noted asAnd
for QC and ASC, the mapping of task i to devices is predetermined, since each container has a defined start and end position on board and on the stack; by fqc(i) And fyc(i) To represent the QC and ASC assigned to task i; the time window allocated to QC task i is then noted asAndsimilarly, the time window assigned to the ASC task j is noted asAnd
3) low-level controller
In this minimum time problem, the AGV is required to complete the operation as quickly as possible while taking into account the stationary obstacle and the moving obstacleAndnumerical method is adopted to calculate the original point state r of the AGV to the container0Transport to target state rfThe shortest time of (d); selecting r from the intersection of QC and ASC for task i0And rf;
During the transportation of task j from the bank side to the stack, the selection is based on QCfqc(i) And ASCfyc(i) Is determined by the task allocation of (c). And after task i is completed, at AGV returns from the stack to the shore side in the process of loading task j (y)ij1) starting and ending points also depend on fqc(i) And fyc(i) In that respect Let T be the width of a given time window. At known time intervals [ 0.,. T-1 ]]Inner, AGVpThe target state r can only be reached at a specific momentfThis is determined by a binary variable at time k. The constraint is as follows:
in the formula, bp(k) E {0,1} is a binary variable, R is a sufficiently large positive real constraint only if bp(k) Equation (26) of 1 is true; the formulas (21) and (27) are restricted when bp(k) When the value is 1, ensuring that the AGV reaches the target position r from the initial state r (k)f;
Assuming that t (k) is a time difference from time 0 to time k (t (k) ═ k), when b (k) ═ 1, (k) b (k) is the completion time of the operation; obtaining the shortest time from the starting point to the end point of the AGV by solving the minimum value of the total completion time; it is also possible to consider improving the energy efficiency of trajectory planning and according to the minimum cost factor λ in the objective functionengMinimizing the sum of the accelerations to achieve a minimum time for the AGV to travel from the start point to the end point; thus, the minimum time optimization problem can be written as follows:
constraint conditions are as follows: (11) (15), (17), (18), (20), (21) and (26), (27).
Wherein u ═ u (0), u (1),.. times, u (T-1)]TRepresenting a continuous decision variable, b ═ bp(0),bp(1),...,bp(T-1)]TA binary decision variable.
The invention has the following effects and advantages:
the invention changes a single shore bridge into a plurality of shore bridges on the basis of generally defined three-stage flow shop scheduling, realizes the operation sequence of related equipment of the mixed flow shop by adopting an integral graph sequence method, and combines an introduced integral graph sequence to plan the collision-free scheduling of all tasks according to a determined layered control structure while considering two AGV collision avoidance constraints, wherein the integral graph sequence simultaneously solves the problems of loading and unloading equipment scheduling and collision-free track planning. The performance and efficiency of handling and transporting containers by the loading and unloading equipment under the condition that AGV collision avoidance constraint is considered in the automatic container terminal are effectively improved.
Drawings
FIG. 1 is a sequence diagram of three equipment shipping containers
FIG. 2 is a schematic diagram of the sequential decision of 6 executing tasks in three stages
FIG. 3 is a schematic diagram of an approximate speed limit for an AGV
FIG. 4 is a schematic diagram of a safety zone of an AGV
FIG. 5 is a schematic view of a static obstacle area
FIG. 6 is a schematic diagram of a hierarchical control structure
FIG. 7 is a schematic diagram of the overall structure of a two-layer controller
TABLE 1 time window for three-stage operation
TABLE 2 selection of start and end points for task i
Detailed Description
Step one, establishing a model of three-stage equipment
In the present invention, consider the multiple QC, multiple AGV and multiple ASC scenarios for an automated container terminal. The operation of three types of equipment can be considered a three stage hybrid flow shop. In a hybrid flow shop, each task performed goes through several stages of processing. Each stage of the same device may process a portion of a task in parallel. The processing sequence of each task is the same and the processing time of each stage of each task is fixed. In the three stage flow shop, a task is defined as a complete process of transporting a container from a ship to a designated location in a heap.
(1) Stage 1: multiple QCs
(2) And (2) stage: multiple AGV
(3) And (3) stage: multiple ASCs
The three-stage hybrid plant is defined with all operations as shown in figure 1. Pi 1Defined as the position of the container i on the ship, Pi 2Anddefined as QC to AGV delivery point location, Pi 3Defined as the location of the delivery point, P, from AGV to ASCi 4Defined as the storage location of the container i in the stacking area.Indicating QC no-load slave Pi 2To Pi 1,Indicating QC full load from Pi 1To Pi 2;Indicating AGV is fully loaded from Pi 2To Pi 3,Indicating AGV is unloaded from Pi 3Return to Pi 2And indicating ASC full load from Pi 3To Pi 4,Indicating ASC idle slave Pi 4Return to Pi 3。
Assuming that there are N containers to be transported from ship to heap, let Φ be the set of tasks (| Φ | ═ N), introduce two virtual job points 0 and N +1, and then define Φ1Φ ∪ {0} and Φ2Φ ∪ { N +1 }. the time constraint for a particular device is as follows:
xij1 indicates that in the first phase, the container task i completes earlier than task j, otherwise, xij=0;
aiIs the completion time of container task i in the first phase; biIs the completion time of the container task i in the second stage; c. CiIs the completion time of the container task i in the third phase; diIs the start time for the transition from task i to task j in the second phase; e.g. of the typeiAGV arrival time at which container i is ready to be transported in the first stage;is thatCompletion time of h1∈{1,2,3},h2E {1,2} represents the operating time window of task i, as shown in Table 1;is the transfer of container i to container j handled by a given AGV in the second stageTime; r is a large positive number; at is the reserved time for two containers to be loaded and unloaded successively by the same ASC.
Inequalities (1) and (4) represent the first task to initialize the QC and AGV, respectively. Inequality (2) represents a relationship between task i and task j processed by QC; inequality (3) ensures that at any moment, the joint point before each QC has at most one AGV; inequalities (5) and (6) represent the start of the second phase after task i is completed in the first phase; inequalities (7) and (8) indicate that task i is started in the third phase after completion of the second phase; inequality (9) represents the relationship between container task j and task i handled by ASC; inequality (10) indicates that the reserved time delta t between two successive tasks is ensured under the constraint of only one AGV at most at the front junction of each heap at any moment; inequalities (11) and (12) represent the transitions of container tasks i and j transported by the AGV.
In accordance with the above constraints, the discrete event dynamics of a three-stage device is considered a hybrid flow shop scheduling problem. These decision variables may be determined by the state monitor by solving a hybrid flow shop scheduling problem. Since the decision variables for the various tasks cannot be determined simultaneously due to collision avoidance considerations, the three-stage hybrid flow shop schedule may be represented using a graphical approach, where each graphical representation processes the corresponding container tasks in a particular order at a particular stage. Integer variable xij,yijAnd zijThe order in which a particular device processes container tasks in the three phases is defined separately. x is the number ofij,yijAnd zijAs shown in fig. 2, an example of 6 container tasks being processed (involving 2QC,3AGV and 3ASC) is given to illustrate the sequence of tasks that each device may process for each phase, where the container tasks being processed ① - ⑥, the virtual container origin taskVirtual container destination task ⑦, with different arrow paths representing different devices as in stage 1, tasks ① and ③ are processed in sequence by one QC and tasks ④ and ⑥ are processed in sequence by another QCThe QCs are processed in sequence, and the first two tasks, task ① and task ④, are processed by the two QCs, and the last two tasks, task ③ and task ⑥, are processed by the two QCs.
TABLE 1 time Window for three-stage operation
The graphical sequence S represents all container tasks that are to be performed at a particular stage of the hybrid flow shop. For phase 1, defineWhere task j is executed after task i. For task i and task j, if task i and task j are accessed sequentially and both task i and task j are executed by a particular device in the graph, then xij1, otherwise xijOne possible graphics order in phase 1 is (①, ④, ④, ④.) for example, task ④ and task ④ are accessed sequentially by the same device, but task ② is not processed after task ④ because task ② and task ④ 1 are not processed by the same device.for a particular phase, there may be multiple graphics orders.for another graphics order for phase 1 may be (④).
In order to schedule all the container tasks of the mixed flow shop in sequence, an overall graph sequence of each stage of the three stages is required, and the index numbers of the arrangement sequence give the operation sequence of the tasks. Therefore, consider introducing a continuous variable bi(i e.di), each task of the overall graph sequence is related to the value of these variables. It will be apparent that the variable values may be arranged in an ascending order to determine the job order of the tasks. And in the three-stage hybrid plant mentioned in step one, if there is R (1-x)ij)+bj>biThen, the sequence S {.. ·, i., j. } (b) }j≥biI, j e Φ, i ≠ j) is a feasible whole graph sequence. The reason is as follows:
considering constraints (7), (8) and (10), there is R (1-z)ij)+bj>dj≥biSo that R (1-z)ij)+bj>bi. Because task i and task j are accessed sequentially for a particular AGV, R (1-y) may be obtainedij)+bj>bi. Thus, for the sequence S {.. ·, i., j. } (b)j≥bi,i,j∈Φ,i≠j),R(1-xij)+bj>bi,R(1-yij)+bj>biAnd R (1-z)ij)+bj>bi. This means that the sequence S is a feasible overall pattern sequence.
Step two, establishing an AGV and a collision avoidance model
1) Establishing AGV dynamic model
Assuming that the dynamic models of each AGV are consistent, all AGVs use the point-quality model to estimate the dynamic behavior of the two-dimensional space
In the formula, AGVpIs at a position ofSpeed isAcceleration of a vehicle(up(k)∈R2),I2Is a 2 × 2 unit array, Δ t is the time interval.
in the formula umaxAnd vmaxUpper limit values for acceleration and velocity, respectively. To simplify the calculations, the constraints on velocity and acceleration both approximate a polygonal constraint using linear equations. The constraints on speed and acceleration are as follows:
wherein M is an arbitrary number. As shown in FIG. 3, a graphical maximum speed approximation of an AGV is presented. The X-axis and Y-axis represent the range of lateral and longitudinal speeds of the AGV, respectively. Circles represent exact constraints, while polygons represent approximate constraints.
2) Establishing static collision avoidance constraint model
Assuming that the AGVs occupy a square safety zone, AGV2, d is the safety distance of a single AGV area, each having a side length of 2d and an area of 2 d. As shown in FIG. 4, a schematic diagram of the safety zone of a single AGV is presented.
As shown in fig. 5, a schematic view of the static obstacle area is given. Assume that FIG. 5(a) is a schematic view of two stationary obstacle areas near a stack, where container 1, AGV2, and landing bridge stationary track 3; the coordinate diagram of the rectangular area of the two tracks of the static obstacle field bridge (ASC) is shown in FIG. 5(b), and the rectangular area can be represented by the coordinates(s) at the lower left cornerlow,x,slow,y) And coordinates of upper right corner(s)high,x,shigh,y) To indicate. To avoid stationary obstacle zones, AGViMust be outside the rectangular area. Therefore, it isThe constraint conditions that satisfy the collision avoidance requirement are as follows:
with binary variables introduced, the constraint (18) can be rewritten to the form of a standard optimization problem as follows:
wherein R is a large positive real number, bin,τ(k) Is a binary variable. Constraints (19) and (20) ensure that at least one of equations (18) is true, thus ensuring that the AGV is outside of the stationary obstacle area.
3) Establishing a moving collision avoidance constraint model
When multiple AGVs transport a container to different destinations, collisions that may occur between the AGVs need to be considered. In each time interval, except for the AGVp1And AGVp2Are different from each other, a minimum safety distance (2d) is maintained between them.
The collision avoidance constraint can be written in the form of:
p2∈[1,...,nagv],p1≠p2in the formula (I), wherein,andrespectively, the k-th step of AGVp1And AGVp2The position of (a). Introducing a binary variable, the constraint (21) can be rewritten as follows:
in the formula, bm,τIs a binary variable, R is a real number, and equations (22) and (23) ensure that equation (21) holds true.
Step three, determining a hierarchical control structure
Based on the dynamic decomposition of the above system, a hierarchical control structure of the container terminal is required to coordinate the phase controllers, as shown in fig. 6, (i.e., high-level controller and low-level controller). The high-level controller comprises a state monitoring controller and a stage controller of each stage. The condition monitoring controller coordinates the three-stage device by determining the sequence and timing of tasks and schedules time windows for each stage's operation. The phase controller assigns a time window for each operation to a particular device. The lower level controllers include respective local controllers. The local controllers are associated with specific devices (QC, AGV and ASC).
According to the hierarchical control structure, the collision-free scheduling of all tasks is planned by the whole graph sequence, and the whole graph sequence simultaneously solves the problems of scheduling of the loading and unloading equipment and planning of collision-free tracks. According to the overall graph sequence, the collision-free scheduling of all tasks is completed sequentially one task after another. For a set of tasks for which an operational sequence is to be determined, the first task planned is used to determine the relevant collision-free trajectory for that task in phase 2, and the relevant operational time for that task in phases 1 and 3, controlled by the higher level controller. After the first task is completed, the high-level controller updates the discrete event dynamics and re-determines the operation sequence and operation time of the remaining tasks. This process is repeated until the scheduling of all tasks is completed. As shown in fig. 7, a schematic diagram of the overall structure of the interaction of the two-layer controller is given.
1) State monitoring controller
The state monitoring controller is used for determining the task sequence of each stage and minimizing the completion time of all tasks of each stage. When the QC and the AGV can be scheduled at any time, the completion time is defined as the completion time of the last ship leaving task. I.e., max { a }1,...,an,e1,...enI.e., | f | ceiling∞,||·||∞Denotes an infinite norm, wherein f ═ a1,a2,...,an,e1,...,en]T。
To determine the order of tasks for all stages of the hybrid flow shop problem, the theorem adds additional constraints. The following definitions are made:
a=[a1,a2,...,an]T
b=[b1,b2,...,bn]T
c=[c1,c2,...,cn]T
d=[d1,d2,...,dn]T
e=[e1,e2,...,en]T
the scheduling problem model can be written as:
constraint conditions are as follows:
and (1) - (12) and associated equation constraints.
When planning the operation of a particular task, the order of operation of the task(s) ((And) Will not change. In particular, it is possible to use, for example,andare interdependent betweenAndstart time between transitionsOrDepending on the interaction between the devices performing task i. This interaction is implemented by a high-level controller for solving the following optimization problem:
constraint conditions are as follows: (1) and (12) and the associated equation constraints. The optimization problem is a linear programming problem, so that the time of the hybrid flow shop can be adjusted quickly.
2) Phase controller
The phase controller assigns each task to a particular device of the respective phase according to the sequence of tasks to be derived. In the hybrid flow shop scheduling problem, this allocation relationship is implicit in the equation constraint, for which the following three mapping functions are defined.
Considering the consistency of the dynamic models of all AGVs in stage 2, an AGV set is defined: Ψagv={1,2,...,nagv}。
fagv:Φ→Ψagv(27)
Wherein f isagvIs to map the task set Φ to the AGV set ΨagvAs a function of (c). f. ofagv(i) The AGV assigned to task i. The time window assigned to AGV task i is noted asAnd
for QC and ASC, the mapping of task i to devices is predetermined, since each container has a defined start and end position on board and on the stack. By fqc(i) And fyc(i) To indicate the QC and ASC assigned to task i. The time window allocated to QC task i is then noted asAndsimilarly, the time window assigned to the ASC task j is noted asAnd
3) low-level controller
By solving a minimum time control problem, AGVAnd the low-level controller realizes collision-free track scheduling of the AGV. In this minimum time problem, the AGV is required to complete the operation as quickly as possible while taking into account the stationary obstacle and the moving obstacleAndnumerical method is adopted to calculate the original point state r of the AGV to the container0Transport to target state rfThe shortest time of (c). Selecting r from the intersection of QC and ASC for task i0And rf。
The exact choice of start and end points for task i is shown in table 2. During the transportation of task j from the bank side to the stack, the selection is based on QCfqc(i) And ASCfyc(i) Is determined by the task allocation of (c). And after task i is completed, in the process that AGV returns from stack to shore side and loads task j (y)ij1) starting and ending points also depend on fqc(i) And fyc(i) In that respect Let T be the width of a given time window. At known time intervals [ 0.,. T-1 ]]Inner, AGVpThe target state r can only be reached at a specific momentfThis is determined by a binary variable at time k. The constraint is as follows:
in the formula, bp(k) E {0,1} is a binary variable, R is a sufficiently large positive real constraint only if bp(k) Equation (28) of 1 is true. Formula (23) and formula (29) are bound when bp(k) When the value is 1, ensuring that the AGV reaches the target position r from the initial state r (k)f。
TABLE 2 selection of start and end points for task i
Assuming that t (k) is a time difference from time 0 to time k (t (k) ═ k), when b (k) ═ 1, t (k) b (k) is the completion time of the operation. And solving the minimum value of the total completion time to obtain the shortest time from the starting point to the end point of the AGV. It is also possible to consider improving the energy efficiency of trajectory planning and according to the minimum cost factor λ in the objective functionengThe sum of the accelerations is minimized to achieve the minimum time for the AGV to travel from the start point to the end point. Thus, the minimum time optimization problem can be written as follows:
constraint conditions are as follows: (13) (17), (19), (20), (22), (23) and (28), (29).
Claims (1)
1. A method for dispatching loading and unloading equipment considering AGV collision avoidance comprises the steps that constraint conditions of the considered collision avoidance comprise static collision and moving collision; the static collision refers to the collision between the AGV and a static obstacle and related equipment near a delivery point in front of the stacking area, and the moving collision refers to the collision between the two AGVs in the process of transporting the container; the scheduling considered includes determining the task order for each particular QC, AGV and ASC and the operating time window for each task; the scheduling is determined by processing a scheduling problem of a hybrid flow shop and an AGV collision avoidance problem, and the collision-free scheduling of a shore bridge, the AGV and a field bridge is realized by combining an integral graph sequence; the method for dispatching the loading and unloading equipment for AGV collision avoidance is characterized by comprising the following steps of:
step one, establishing a model of three-stage equipment
In the three-stage flow shop, a task is defined as a complete process of transporting a container from a ship to a specified position in a stacking area;
(1) stage 1: multiple QCs
(2) And (2) stage: multiple AGV
(3) And (3) stage: multiple ASCs
Assuming that there are N containers to be transported from ship to heap, let Φ be the set of tasks (| Φ | ═ N), introduce two virtual job points 0 and N +1, and then define Φ1Φ ∪ {0} and Φ2Phi ∪ { N +1}, the time constraint for a particular device is as follows:
xij1 indicates that in the first phase, the container task i completes earlier than task j, otherwise, xij=0;
yij1 indicates that in the second phase, the container task i completes earlier than task j, otherwise, yij=0;
zij1 indicates that in the third phase, the container task i completes earlier than task j, otherwise zij=0;
aiIs the completion time of container task i in the first phase; biIs the completion time of the container task i in the second stage; c. CiIs the completion time of the container task i in the third phase; diIs the start time for the transition from task i to task j in the second phase; e.g. of the typeiIs the AGV arrival time of the first stage ready to transport container i;Is thatCompletion time of h1∈{1,2,3},h2∈{1,2};Represents the corresponding container operation, wherein h1∈{1,2,3},h2∈{1,2},Indicating QC no-load slave Pi 2To Pi 1,Indicating QC full load from Pi 1To Pi 2;Indicating AGV is fully loaded from Pi 2To Pi 3,Indicating AGV is unloaded from Pi 3Return to Pi 2And indicating ASC full load from Pi 3To Pi 4,Indicating ASC idle slave Pi 4Return to Pi 3;Is the transfer time of container i to container j handled by the designated AGV in the second phase; r is a large positive number; Δ t is the reserved time for two containers to be loaded and unloaded successively by the same ASC;
inequalities (1) and (4) represent the first task of initializing the QC and the AGV, respectively; inequality (2) represents a relationship between task i and task j processed by QC; inequality (3) ensures that at any moment, the joint point before each QC has at most one AGV; inequalities (5) and (6) represent the start of the second phase after task i is completed in the first phase; inequalities (7) and (8) indicate that task i is started in the third phase after completion of the second phase; inequality (9) represents the relationship between container task j and task i handled by ASC; inequality (10) indicates that the reserved time delta t between two successive tasks is ensured under the constraint of only one AGV at most at the front junction of each heap at any moment; inequalities (11) and (12) represent the transitions of container tasks i and j transported by the AGV;
graphically representing the three-stage hybrid flow shop schedule, wherein each graphical representation processes corresponding container tasks in a particular order at a particular stage; integer variable xij,yijAnd zijRespectively defining the sequence of the container tasks processed by specific equipment in the three stages; x is the number ofij,yijAnd zijThe values of (a) may be represented by a graph of the stages;
the graphic sequence S represents the scheduling sequence of all container tasks to be executed in a specific stage of the hybrid flow shop; for phase 1, defineWherein task j is executed after task i; for task i and task j, if task i and task j are accessed sequentially and both task i and task j are executed by a particular device in the graph, then xij1, otherwise xij0; stages 2 and 3 are defined similarly;
taking into account the introduction of a continuous variable bi(iE Φ), each task of the overall graph sequence is related to the values of these variables; obviously, in a three stage hybrid plant, if R (1-x) is presentij)+bj>biThen, the sequence S {.. ·, i., j. } (b) }j≥biI, j ∈ Φ, i ≠ j) is a feasible overall graph sequence; and the variable values can be arranged according to an ascending method to determine the operation sequence of the tasks;
step two, establishing an AGV and a collision avoidance model
1) Establishing AGV dynamic model
In the formula, AGVpIs at a position ofSpeed isAcceleration of a vehicle(up(k)∈R2),I2Is a 2 x 2 unit array, Δ T is the reserved time to load and unload two containers in succession by the same ASC, T is the width of a given time window, i.e. assuming a known time interval [0];
in the formula umaxAnd vmaxUpper limit values of acceleration and velocity, respectively, M being an arbitrary number;
2) establishing static collision avoidance constraint model
The rectangular area may be represented by the lower left corner coordinate(s)low,x,slow,y) And coordinates of upper right corner(s)high,x,shigh,y) To represent; to avoid the static obstacle area, the position of AGVi must be outside the rectangular area; therefore, the constraint conditions for satisfying the collision avoidance requirement are as follows:
with binary variables introduced, constraint (16) can be rewritten into the form of a standard optimization problem as follows:
in the formula, d is the safety distance of a single AGV region, the side length of the region is 2d, and the area of the region is 2d multiplied by 2 d; r is a very large positive real number, bin,τ(k) Is a binary variable; constraints (17) and (18) ensure that at least one of equations (16) is true;
3) establishing a moving collision avoidance constraint model
In each time interval, except for the AGVp1And AGVp2Are different from each other,a minimum safety distance 2d is kept between the two; the collision avoidance constraint can be written in the form of:
p2∈[1,...,nagv],p1≠p2in the formulaAndp of AGV in k1And p2The position of (a); introducing a binary variable, the constraint (19) can be rewritten as follows:
in the formula, bm,τIs a binary variable, R is a sufficiently large positive real number, and equations (20) and (21) ensure that equation (19) holds;
step three, determining a hierarchical control structure
According to the hierarchical control structure, the collision-free scheduling of all tasks is planned by the whole graph sequence, and the whole graph sequence simultaneously solves the problem of collision-free track planning and the scheduling of loading and unloading equipment; according to the overall graph sequence, sequentially completing collision-free scheduling of all tasks one by one; for a group of tasks with an operation sequence to be determined, the first planned task is used for determining the relevant collision-free track of the task in the phase 2 and the relevant operation time of the task in the phase 1 and the phase 3 controlled by a high-level controller; after the first task is completed, the high-level controller updates the discrete event dynamic and determines the operation sequence and the operation time of the rest tasks again; this process is repeated until the scheduling of all tasks is completed;
1) state monitoring controller
The state monitoring controller is used for determining the task sequence of each stage and minimizing the completion time of all tasks of each stage; when the QC and the AGV can be scheduled at any time, the completion time is defined as the completion time of the last ship leaving task; i.e., max { a }1,...,an,e1,...enI.e., | f | ceiling∞,||·||∞Denotes an infinite norm, wherein f ═ a1,a2,...,an,e1,...,en]T;
In order to determine the task order of all stages of the mixed flow shop problem, the theorem needs to add additional constraints; the following definitions are first made:
a=[a1,a2,...,an]T
b=[b1,b2,...,bn]T
c=[c1,c2,...,cn]T
d=[d1,d2,...,dn]T
e=[e1,e2,...,en]T
the scheduling problem model can be written as:
constraint conditions are as follows:
and (1) - (12) and associated equation constraints;
betweenAndstart time between transitionsOrDependent on the interaction between the devices performing task i; this interaction is implemented by a high-level controller that solves the following optimization problem:
constraint conditions are as follows: (1) (12) and associated equality constraints; the optimization problem is a linear programming problem, so that the time of the hybrid flow shop can be quickly adjusted;
2) phase controller
Considering the consistency of the dynamic models of all AGVs in stage 2, an AGV set is defined: Ψagv={1,2,...,nagv}
fagv:Φ→Ψagv, (25)
Wherein f isagvIs to map the task set Φ to the AGV set ΨagvA function of (a); f. ofagv(i) An AGV assigned to task i; the time window assigned to AGV task i is noted asAnd
for QC and ASC, use fqc(i) And fyc(i) To represent the QC and ASC assigned to task i; the time window allocated to QC task i is then noted asAndsimilarly, the time window assigned to the ASC task j is noted asAnd
3) low-level controller
In this minimum time problem, the AGV is required to complete the operation as quickly as possible while taking into account the stationary obstacle and the moving obstacleAndnumerical method is adopted to calculate the original point state r of the AGV to the container0Transport to target state rfThe shortest time of (d); selecting r from the intersection of QC and ASC for task i0And rf;
The selection is based on QCf during the course of task j being shipped from the bank side to the stackqc(i) And ASCfyc(i) Is determined by the task allocation of (c); and after task i is completed, in the process that AGV returns from stack to shore side and loads task j (y)ij1), start and endThe point also depends on fqc(i) And fyc(i) (ii) a Assuming that T is the width of a given time window, T-1 is applied at known time intervals [0]Inner, AGVpThe target state r can only be reached at a specific momentfThis is determined by a binary variable at time k; the constraint is as follows:
in the formula, bp(k) E {0,1} is a binary variable, R is a sufficiently large positive real number, constrained only if bp(k) Equation (26) of 1 is true; the formulas (21) and (27) are restricted when bp(k) When the value is 1, ensuring that the AGV reaches the target position r from the initial state r (k)f;
Assuming that t (k) is a time difference from time 0 to time k (t (k) ═ k), when b (k) ═ 1, (k) b (k) is the completion time of the operation; obtaining the shortest time from the starting point to the end point of the AGV by solving the minimum value of the total completion time; it is also possible to consider improving the energy efficiency of trajectory planning and according to the minimum cost factor λ in the objective functionengMinimizing the sum of the accelerations to achieve a minimum time for the AGV to travel from the start point to the end point; thus, the minimum time optimization problem can be written as follows:
constraint conditions are as follows: (11) - (15), (17), (18), (20), (21) and (26), (27);
wherein u ═ u (0), u (1),.. times, u (T-1)]TRepresenting a continuous decision variable; b ═ bp(0),bp(1),...,bp(T-1)]TRepresenting binary blocksA policy variable.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201710984769.5A CN107678433B (en) | 2017-10-20 | 2017-10-20 | Loading and unloading equipment scheduling method considering AGV collision avoidance |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201710984769.5A CN107678433B (en) | 2017-10-20 | 2017-10-20 | Loading and unloading equipment scheduling method considering AGV collision avoidance |
Publications (2)
Publication Number | Publication Date |
---|---|
CN107678433A CN107678433A (en) | 2018-02-09 |
CN107678433B true CN107678433B (en) | 2020-05-29 |
Family
ID=61140320
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201710984769.5A Active CN107678433B (en) | 2017-10-20 | 2017-10-20 | Loading and unloading equipment scheduling method considering AGV collision avoidance |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN107678433B (en) |
Families Citing this family (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN108596519B (en) * | 2018-05-14 | 2021-06-04 | 大连海事大学 | Automatic wharf ALV optimal scheduling method |
CN108983779B (en) * | 2018-07-24 | 2021-12-21 | 合肥哈工库讯智能科技有限公司 | AGV trolley traffic control regulation and control method based on path analysis |
CN109048909B (en) * | 2018-08-29 | 2020-12-18 | 广州市君望机器人自动化有限公司 | Branch node type path scheduling method and device, background server and first robot |
CN110516976B (en) * | 2019-08-30 | 2022-12-13 | 大连海事大学 | Automatic container terminal AGV (automatic guided vehicle) scheduling method capable of selecting loading and unloading completion time |
CN110683463B (en) * | 2019-10-17 | 2020-09-04 | 上海海事大学 | Real-time scheduling-control cascade system and method for double cranes |
CN113435722B (en) * | 2021-06-17 | 2024-01-23 | 上海海事大学 | Multi-device mixed scheduling method for U-shaped automatic wharf and electronic device |
CN114371612A (en) * | 2021-12-14 | 2022-04-19 | 珠海格力智能装备有限公司 | Moving object acceleration control method, moving object acceleration control device, computer equipment and storage medium |
Family Cites Families (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US6092010A (en) * | 1997-09-03 | 2000-07-18 | Jervis B. Webb Company | Method and system for describing, generating and checking non-wire guidepaths for automatic guided vehicles |
CN105740979B (en) * | 2016-01-29 | 2020-02-14 | 上海海事大学 | Intelligent dispatching system and method for multiple automatic guided vehicles of automatic wharf |
CN107092263A (en) * | 2017-06-21 | 2017-08-25 | 安徽瑞弋自动化科技有限公司 | A kind of AGV vehicle dispatchings system and method |
CN107169717A (en) * | 2017-06-26 | 2017-09-15 | 海航创新科技研究有限公司 | Logistics transportation methods, devices and systems based on cloud platform |
CN107678279A (en) * | 2017-10-09 | 2018-02-09 | 上海海事大学 | A kind of AGV method for planning track for considering to evade constraint |
-
2017
- 2017-10-20 CN CN201710984769.5A patent/CN107678433B/en active Active
Also Published As
Publication number | Publication date |
---|---|
CN107678433A (en) | 2018-02-09 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN107678433B (en) | Loading and unloading equipment scheduling method considering AGV collision avoidance | |
Xin et al. | Energy-aware control for automated container terminals using integrated flow shop scheduling and optimal control | |
Xin et al. | Control of interacting machines in automated container terminals using a sequential planning approach for collision avoidance | |
Kim et al. | An optimal routing algorithm for a transfer crane in port container terminals | |
CN106651049B (en) | Rescheduling method for automatic container terminal loading and unloading equipment | |
KR102015006B1 (en) | System and method for controlling at least two automated non-passing rail mounted gantry cranes | |
Fottner et al. | Autonomous systems in intralogistics–state of the art and future research challenges | |
CN114023094B (en) | Wharf vehicle scheduling system, method, equipment and medium | |
CN106773686B (en) | Path model method for building up is dispatched with piler under the double vehicle operational modes of rail | |
CN110716452A (en) | Dispatching simulation method based on intelligent stereoscopic warehouse of multilayer shuttle | |
CN111573114A (en) | Storage container transport control method, storage container transport control device, storage container transport control apparatus, and storage medium | |
Walenta et al. | A decentralised system approach for controlling AGVs with ROS | |
Hu et al. | Multi-AGV dispatching and routing problem based on a three-stage decomposition method | |
WO2015011661A2 (en) | Device and method for optimising the movement of automated-guided vehicles, and the like | |
Stahlbock et al. | Efficiency considerations for sequencing and scheduling of double-rail-mounted gantry cranes at maritime container terminals | |
CN111781927A (en) | Scheduling and distributing method for multi-robot cooperative transportation task | |
CN115140657A (en) | Multi-crown-block scheduling and collision avoidance method and system | |
Solichudin et al. | Conflict-free dynamic route multi-agv using dijkstra Floyd-warshall hybrid algorithm with time windows | |
Nguyen et al. | Heuristic algorithms for constructing transporter pools in container terminals | |
CN113743747A (en) | Multi-AGV cooperative scheduling method and device in workshop environment | |
Dkhil et al. | Optimization of container handling systems in automated maritime terminal | |
Chaikovskaia et al. | Sizing of a fleet of cooperative robots for the transport of homogeneous loads | |
Chaikovskaia et al. | Sizing of a fleet of cooperative and reconfigurable robots for the transport of heterogeneous loads | |
Ahmed et al. | Improving productivity of yard trucks in port container terminal using computer simulation | |
Deja et al. | Simulation studies into quayside transport and storage yard operations in container terminals |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |